Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix memset size #9840

Merged
merged 1 commit into from
Nov 23, 2021
Merged

Fix memset size #9840

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion onnxruntime/contrib_ops/cuda/inverse.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,7 +150,7 @@ Status Inverse::ComputeInternal(OpKernelContext* ctx) const {
}

IAllocatorUniquePtr<int> info = GetScratchBuffer<int>(num_batches);
CUDA_RETURN_IF_ERROR(cudaMemsetAsync(info.get(), 0, num_batches, Stream()));
CUDA_RETURN_IF_ERROR(cudaMemsetAsync(info.get(), 0, num_batches * sizeof(int), Stream()));
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just curious, when will this async op be synced ?

Copy link
Member

@hariharans29 hariharans29 Nov 23, 2021 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It doesn't need explicit syncing right ? When launching operations using a single stream (per-session compute stream : Stream()), all CUDA operations in that stream are serialized and so they execute in the order in which they are queued. Anything queued after this on the same stream will execute only after this operation.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, Hari is correct.

IAllocatorUniquePtr<int> pivots = GetScratchBuffer<int>(rows * num_batches);

utils::MLTypeCallDispatcher<float, double, MLFloat16> t_disp(input->GetElementType());
Expand Down